let’s work it out (or we’ll see you in court): litigation
TRANSCRIPT
Munich Personal RePEc Archive
Let’s work it out (or we’ll see you incourt): litigation and private disputeresolution in vertical exchangerelationships
Lumineau, Fabrice and Oxley, Joanne
2012
Online at https://mpra.ub.uni-muenchen.de/38361/
MPRA Paper No. 38361, posted 25 Apr 2012 12:32 UTC
Electronic copy available at: http://ssrn.com/abstract=1816402
1
LET’S WORK IT OUT (OR WE’LL SEE YOU IN COURT…) Litigation and Private Dispute Resolution in Vertical Exchange Relationships
Fabrice Lumineau
University of Technology Sydney
P.O. Box 123
Broadway NSW 2007
Australia
Joanne E. Oxley
Rotman School of Management
University of Toronto
105 St. George Street
Toronto ON M5S 3E6
Canada
Forthcoming, Organization Science
Abstract:
We examine how partners in vertical exchange relationships actually resolve disputes that are sufficiently
serious to get lawyers involved. Reaching beyond the usual domain of organizational and management research
we leverage findings from law and economics to offer a novel organizational perspective on litigation and
private dispute resolution, and develop hypotheses about the likelihood of litigation in different exchange
settings. Our empirical analysis generates three sets of new findings: First, counter to the received wisdom we
see that the involvement of lawyers does not necessarily signal the bitter end of an exchange relationship, as
firms frequently manage to avoid litigation and resolve their disputes privately, and do so in a manner that
accords with our theoretical predictions. Second, we see that familiarity with exchange partners does not
automatically lead to increased willingness to work things out: rather, our empirical results suggest that the
impact of exchange duration on parties’ willingness to resolve disputes privately is contingent on the development of norms of cooperation; in the event that such norms do not develop, the probability of a
litigated outcome actually increases over time. Finally, we see that firms’ willingness to work things out privately is also influenced positively by the shadow of the future. These findings are suggestive of a “discriminating alignment” between exchange characteristics and the choice of dispute resolution procedure, and thus inject
important new evidence into ongoing discussions about the legal underpinnings of different governance forms.
Key words: contracts; dispute resolution; litigation.
Electronic copy available at: http://ssrn.com/abstract=1816402
2
Introduction
The organizational and relational characteristics supporting the development of cooperative norms and
commitment among exchange partners is a topic of enduring interest among organizational scholars. With few
exceptions (notably Ariño and de la Torre 1998, Park and Ungson 2001) research to date has examined
commitment and performance in ongoing stable exchange relationships (e.g., Gulati 1995, Dyer and Singh 1998,
Poppo and Zenger 2002). But what happens when exchange relationships run into difficulty, disputes escalate,
and lawyers get involved? Are such relationships inevitably headed towards litigation, as one might presume, or
are firms still able to work things out among themselves? How does the likelihood of litigation vary with
characteristics of the current exchange, historical relationship, or expected future exchange?
In this paper we employ a unique dataset containing detailed information on contract disputes handled
by one European law firm to examine how partners in vertical exchange relationships actually resolve disputes
that are sufficiently serious to get lawyers involved. This is an under-researched but important aspect of inter-
firm governance: while much of the organizational literature on inter-firm exchange and alliances has the threat
of litigation in the background, there is almost no empirical research investigating the circumstances leading to
litigation. At the same time, as detailed in Williamson’s (1991) seminal paper, differences in legal supports –
particularly as they relate to dispute resolution – are believed to constitute an important distinguishing feature
of alternative governance arrangements. Gaining a better understanding of how firms choose between litigation
and private dispute resolution procedures (negotiation, mediation and arbitration) thus has the potential to
contribute significantly to our understanding of the governance of vertical exchange relationships and other
‘hybrid’ organizations.
Our analysis focuses on several important dimensions of vertical exchange relationships identified in the
management literature, including the technical complexity of the focal exchange, the firms’ history of prior
exchange, (the so-called “shadow of the past”) and expectations about future exchange (the “shadow of the
future”). In developing our hypotheses linking these characteristics to the likelihood of litigation we also draw on
3
an established theoretical literature in law and economics on the settlement of contract disputes (Bebchuk
1984, Johnson et al. 2002, Richman 2004). This literature has been largely ignored by organizational researchers,
perhaps because legal scholars have focused primarily on variation in litigation rates across broad classes of
cases, and thus have not engaged the kinds of organizational or relational characteristics at the center of related
studies in the management domain. By bringing these literatures together we offer a novel organizational
perspective on litigation and private dispute resolution, and develop hypotheses about the likelihood of
litigation in different exchange settings. Our research thus gives us a window on dispute resolution in exchange
relationships that are under significant duress, something that represents an important extension of both
organizational research and related research in law and economics.
The remainder of the paper is organized as follows: We first lay out the theoretical arguments in Section
1, focusing first on the basic attributes of different dispute resolution modes emphasized by legal scholars and
then developing our own organizational perspective on litigation versus private dispute resolution. The empirical
data and methods are introduced in Section 2; results follow in Section 3. Section 4 discusses our findings and
concludes.
1. Litigation and Private Dispute Resolution
1.1. The law and economics view
Despite the best efforts of managers and lawyers to guard against future contingencies that could
threaten cooperation, many inter-firm exchange relationships fall into dispute (Mohr and Spekman 1994). And
while most such disputes are handled through informal discussions or routine procedures, in more serious cases
advice may be sought from lawyers as managers explore ways to resolve the conflict, including possibly through
litigation. It is important to note, however, that the involvement of lawyers does not automatically mean that a
dispute will end up in court. Indeed, contrary to popular belief, lawyers often play an important role in
facilitating private resolution of serious disputes in commercial settings (Riskin et al. 2009). Greater distance and
objectivity means that lawyers are often able to uncover potential zones of agreement and thus enhance the
4
probability that disputing parties can reach a negotiated settlement – either independently, or through
mediation involving a neutral third party.
As emphasized in the law and economics literature, a significant reason why lawyers typically counsel
their clients to pursue private dispute resolution is to avoid the large fixed costs associated with litigation – not
only the direct costs of lawyers’ and executives’ time during the pre-trial discovery phase and the trial
proceedings themselves, but also indirect costs stemming from the delays that are endemic to court
proceedings (Bebchuk 1984). These fixed costs create a “bargaining surplus” that can be shared by the parties if
a negotiated settlement is achieved and the dispute is terminated before it goes to trial (Gould 1973). Thus, for
example, in a high-profile dispute between Ford Motors and Navistar (Ford’s supplier of diesel engines for over
thirty years) while lawyers got involved early in the dispute, and litigation was initiated, ongoing negotiations led
to a private settlement well in advance of any potential court ruling, and at significant lower cost to the
disputing parties (Simonin 2007).
In addition to cost savings, recent legal research has also highlighted other advantages of private dispute
resolution procedures. Johnson et al. (2002, p. 229), for example, argue that negotiated or mediated dispute
settlements are frequently superior to litigated outcomes because, inter alia, “…they can consider information
that cannot be introduced in court, such as impressionistic evidence about business trends or judgments about
the quality of items sold. They can base their decisions on a firm’s behavior over time, on probabilistic patterns
that would not be admissible evidence in court.”
Given the dual features of greater nuance and lower cost associated with private dispute resolution
procedures, one might well ask why a commercial dispute ever ends in litigation. Contemplating this apparent
paradox, legal scholars have argued that litigation will only take place when a fundamental disagreement
develops between the parties over who will likely prevail in a trial, and/or what judgment will be handed down,
and this disparity effectively reduces the perceived bargaining surplus to zero. In theoretical models, reasons for
such disagreements have typically been ascribed to information asymmetries between the exchange partners –
5
and thus uncertainty about the actual facts of a case – or uncertainty about the legal standards or decision rules
that will be applied by the judge or jury (Bebchuk 1984, Priest and Klein 1984). In empirical investigations, legal
scholars have primarily explored the effect of uncertain legal standards, comparing the proportion of lawsuits
that are settled privately across broad classes of cases – torts, civil or prisoner rights, contracts, intellectual
property, labor disputes, etc. (see Siegelman and Waldfogel 1999 for a summary of the evidence). These studies
have found broad support for the notion that uncertainty about legal standards increases the number of cases
filed with the court that actually go to trial; they unfortunately do little to illuminate the link between litigation
and features of a particular inter-firm exchange relationship, the focus of our own inquiry and the issue of most
interest to organizational scholars. For this we turn to the organizations literature on sources of information
asymmetry and uncertainty in exchange relationships.
1.2. The organizational view – exchange and relational characteristics and the likelihood of litigation
Uncertainty has been a focal concept in theories of organization, at least since Knight (1921). And while
researchers have explored a variety of sources and implications of uncertainty (Daft and Macintosh 1981, Carson
et al. 2006), we restrict our attention here to two types of uncertainty that we believe are particularly relevant
to firms’ abilities to predict the likely outcome of litigation, which in turn impacts their willingness to negotiate a
settlement “in the shadow of the court,” as discussed above: these are ‘technical uncertainty’ related to the
complexity of the focal exchange (Simon 1962) and ‘behavioral uncertainty’ related to actions and motivations
of the exchange partner (Williamson 1985).1 For behavioral uncertainty, we focus in particular on how firms’
willingness to work things out privately changes as behavioral uncertainty evolves over the course of an
exchange relationship.
1 Other sources of uncertainty identified in the organizations literature, such as volatility in the exchange environment (e.g.,
Carson et al. 2006) are more likely to impact the frequency of disputes then the ability to predict litigation outcomes (and
thus the likelihood that a disputes is settled prior to going to court).
6
Prior research in organizational economics suggests that technical uncertainty, rooted in humans’
limited understanding of nature (Slater and Spencer 2000) is increasing in the complexity of the technology
involved in an exchange because technical “glitches” become more likely as complexity increases (Hoopes and
Postrel 1999), and it becomes increasingly difficult to demonstrate causal links between exchange partners’
efforts and observed outcomes (Monteverde and Teece 1982). Thus, starting from a baseline of technically
simple transactions (such as supply contracts for simple standard parts requiring little customization), we should
expect that as technical complexity increases, the likelihood of divergent expectations regarding litigation
outcomes will also increase. As a result, while firms still have an incentive to settle disputes involving complex
transactions privately to save the fixed costs of litigation, they and their lawyers are more likely to make
mistakes in their assessments of what constitutes an appropriate settlement, thereby ending up in court with
greater frequency. This implies that we will see a positive relationship between technical complexity and the
likelihood of litigation.
One important caveat to the above argument, however – and indeed to the related research linking
uncertainty to the likelihood of litigation in law and economics – is that it focuses exclusively on the challenges
that disputants (managers and their lawyers) face in making attributions and estimating costs associated with
particular operational failures. However, for highly complex transactions, even if the firms involved in the
exchange have a reasonable understanding of the facts of the case, there may be significant limitations on the
verifiability of these facts within the constraints of judicial proceedings (Grossman and Hart 1986). Williamson
(1991, pp. 272–73) argues that under these circumstances, parties recognize the limits of court ordering and
move towards a “neoclassical” contracting regime (Llewellyn 1931), where court ordering is explicitly rejected in
favor of private dispute resolution – negotiation, mediation, or arbitration.
Together these arguments imply a non-linear relationship between technical uncertainty and the
likelihood of litigation: for disputes involving technically very simple exchanges we should expect that litigation
will be rare, as parties can reach agreement relatively easily and, by settling privately, can avoid the costs of
7
litigation; indeed at the limit we would expect managers to settle problems informally so that such cases would
not to come to the attention of lawyers at all. With increasing technical uncertainty, misattributions and errors
in managers’ (and lawyers) predictions of likely litigation outcomes will occur with increasing frequency,
undermining efforts to reach agreement and so landing the disputing parties in court. However, for transactions
of very high technical complexity, we expect that litigation will again be rare: we should expect firms to
recognize the limits of the court to handle highly technically-complex disputes – as well as the likelihood that
they themselves may make misattributions and errors in predicting court outcomes – and will thus commit to
private dispute resolution procedures that support the nuanced understanding required for settlement of these
cases. It is in the middle range of complexity where litigation is likely to be more common then, as firms may
underestimate the difficulties associated with private settlement and/or overestimate the ability of the court to
handle the dispute. This combined argument thus leads to the following hypothesis:
H1: There is an inverted U-shaped relationship between the technical complexity of the exchange and the
likelihood of litigation.
Behavioral uncertainty refers to uncertainty about others’ future actions and motivations and thus the
likelihood of cooperation versus opportunism within exchange relationships. Such uncertainty stems from the
observation that “some individuals are opportunistic some of the time, and differential trustworthiness is rarely
transparent ex ante” (Williamson 1985, p. 64). Economic models of repeated exchange focus on active selection
of cooperative partners when new contracts are awarded or contracts are renewed. As long as exchange
partners cooperate, exchange continues, contracts are renewed, etc.; if one partner deviates and acts
opportunistically, exchange is terminated (Klein 1996). Based on this same idea, firms are sometimes advised to
test new exchange partners in relatively simple short-term projects so that uncooperative partners can
effectively be weeded out prior to making any longer-term or high-volume exchange commitments (Liker and
Choi 2004).
8
If firms systematically condition contract renewal on the basis of cooperative behavior in prior
interactions, then in a cross-section we should observe a positive correlation between the total elapsed duration
of an exchange relationship and the average perceived level of cooperation between partners. This in turn
should lead to fewer disputes emerging in exchange relationships of longer elapsed duration and a higher
willingness to work things out in the event that a serious dispute does emerge.
A similar conclusion can be drawn from research that focuses not on weeding out of uncooperative
exchange partners, but rather on inter-partner learning: it has been suggested that knowledge of an exchange
partner’s capabilities and constraints increases over time, as does understanding of the technical and exchange
domain (Zollo et al. 2002, Argyres et al. 2007).2 In this view, learning predictably decreases uncertainty and
information asymmetries among exchange partners over time, reinforcing the positive selection effects
emphasized by Klein (1996) and others. Thus we have two paths to the following hypothesis:
H2: The likelihood of litigation is decreasing in the elapsed duration of the exchange relationship at the
time the dispute arises.
An alternative perspective on the emergence of cooperative norms in exchange relationships takes a
less calculative view than that discussed above and instead emphasizes powerful cognitive and emotional
human traits (Lewis and Weigert 1985). In this view repeated interactions over the course of exchange lead to
the spontaneous emergence of “familiarity-based trust,” as personal ties between members of the two
organizations develop, leading to feelings of affiliation (Gulati 1995), “pressures for conformity to expectations”
(Macaulay 1963, p. 63) and “fine-grained information transfer, and joint problem-solving” (Uzzi 1996, p. 677).
These features are indicative of an increase in the willingness and ability of partners in repeated exchange to
2 We should note that learning is logically connected to the intensity or volume of the exchange as well as the duration of
exchange. Unfortunately we do not have reliable measures of exchange volume in our data and so we focus solely on
duration. However, based on prior evidence it is likely that duration and volume of exchanges are highly correlated (see,
e.g., Vanneste and Puranam 2010).
9
work things out and maintain relationship stability, reinforcing the association between relationship duration
and private dispute resolution discussed above.
An active recent literature has nonetheless begun to challenge the notion that a longer history of
exchange automatically leads to the development of trust and cooperative norms (See, e.g., Anderson and Jap
2005, Poppo et al. 2008). Some writers on collaborative strategies have noted the existence of an initial
“honeymoon period,” where exchange partners may give each other the benefit of the doubt in the early stages
of the relationship (Kanter 1994). In this view, it is only when the relationship inevitably faces its first significant
‘shock’ (e.g., due to changes in the environment or the strategic context) that attitudes towards the exchange
partner – positive or negative – begin to solidify (Ariño and de la Torre 1998, Gulati and Sytch 2008). Similarly,
Poppo et al. (2008, p. 1200) note the tentative nature of behavioral attributions in the early stages of an
exchange relationship: “Trust, if it exists, likely is very fragile because little personal history provides a basis for
it.” This argument implies that the effect of elapsed duration on firms’ willingness to work things out will be
significantly muted until and unless norms of cooperation develop in the exchange relationship. However,
subsequent to the development of such norms we would expect that the likelihood of litigation will continue to
decrease over time as firms continue to learn from each other, so reducing information asymmetries between
the parties, decreasing uncertainty over litigation outcomes, and facilitating negotiation and private dispute
resolution. Thus we suggest:
H3: The negative effect of elapsed duration on the likelihood of litigation is weakened if cooperative
norms are absent from the exchange relationship.
In addition to the impact of the history of exchange on dispute resolution, prior research also suggests
that dispute resolution is shaped by the “shadow of the future” – when firms have the expectation of continued
profitable exchange into the future, their incentives to resolve disputes to preserve the exchange relationship
are predictably increased. This proposition is derived from well-established economic models of repeated
exchange (Axelrod 1984) and is closely related to the selection models discussed above (e.g., Klein 1996): in
10
these models it is precisely the threat of contract termination (or non-renewal), and loss of future rents, that
brings discipline into an exchange relationship and supports the development of cooperative norms among
exchange partners. Moreover, the longer the expected time horizon over which future benefits are expected to
accrue, the higher are the benefits of continued cooperation.
One implication of this logic is that firms are less likely to act opportunistically when the shadow of the
future is long – a proposition that has found significant support in prior empirical studies of inter-organizational
relationships (e.g., Heide and Miner 1992). An additional implication, however, is that when the shadow of the
future is long firms may be loathe to forego the expected gains from future cooperation even if they suspect that
their exchange partner may have acted opportunistically (Farrell and Maskin 1989). Dispute resolution in this
context becomes more complex as firms seek to balance the need to preserve discipline within the exchange (by
punishing defection) with the desire to maintain the relationship and reap the benefits of continued exchange.
As discussed earlier, courts are particularly ill-suited to making the kinds of nuanced judgments that are tailored
to the needs of the parties in such particular circumstances; as such, we would expect firms to eschew litigation
and instead opt for private dispute resolution in these cases. This reasoning thus leads to our final hypothesis:
H4: The likelihood of litigation is decreasing in the expected future duration of the exchange relationship
at the time the dispute arises.
2. Empirical Data and Methods
2.1. Empirical Context and Sample Description
Our empirical study employs a sample of contract disputes arising in vertical exchange relationships. The
first author was granted access by a French law firm to all legal files concerning contract disputes handled by the
11
firm between 1991 and 2005.3 Data collection took place over a four-month period during which time the
researcher was able to gain insights into the legal context and the practices and routines of the law firm through
daily informal conversations with lawyers and administrative staff. Additional interviews were conducted with
lawyers unrelated to this law firm, and with law professors specialized in contract law. The data collection
process yielded information on disputes related to 102 vertical exchange contracts involving 178 unique firms.
Data obtained from the legal files include all documents issued by each party to the contract and
exchanged during the dispute resolution process, as well as additional information requested by the lawyers
from the client, such as the initial context of the contracting relationship, the origin of the conflict, and its
evolution prior to the law firm getting involved. Due to the highly confidential nature of the data the researchers
were not able to speak directly to the firms involved in the disputes, nor is it possible to identify the companies
in the sample by name.
While we believe that we are the first management scholars to collect dispute data from lawyers’ notes,
our use of third-party notes as a source of data in management research is far from unique (see, e.g., Klein
Woolthuis et al. 2005, Bouwens and van Lent 2006). Lawyers’ notes have some attractive features as a data
source for a study of dispute resolution as they provide a near-real time record of firms’ perceptions and
intentions as the dispute emerged and progressed. Lawyers’ notes are usually compiled during initial meetings
between the lawyer assigned to the case and the client – typically the CEO in a small firm, or the mid-level
manager responsible for the exchange relationship in a larger firm. Moreover, because it is important for
lawyers to obtain an accurate picture of past events and future intentions, they use a variety of interviewing
strategies to get to the facts and to minimize exaggeration and obfuscation by managers who may be
emotionally involved and otherwise apt to portray their firm in an overly positive light (Tractenberg 1984). These
3 Discussions with lawyers at the firm indicated that they consider their firm to be quite representative of mid-size
generalist corporate law firms in France. The firm’s generalist status is also reflected in the wide variety of clients represented in terms of size and industry focus.
12
interviews are confidential, and views expressed are not subject to legal disclosure, nor are they shared with
lawyers representing the other firm. All in all, it seems reasonable to expect that many of the problems of over-
simplification, faulty post hoc attributions and other retrospective biases or simple lapses of memory that
plague first-person retrospective accounts may be reduced in these third-party notes (Golden 1992).
A significant limitation of our data source, however, is that we only observe disputes when lawyers
become involved and disputes resolved prior to lawyer involvement will be missing entirely from our data set.
This may introduce problematic selection bias into our empirical analysis; we therefore explore possible
selection effects in the robustness section following presentation of the main results. At the same time, we also
note that our data set is much less restrictive than those used in previous empirical studies of settlement and
litigation in the law and economics literatures, which rely almost exclusively on samples of cases actually filed
with the court (see, e.g., Siegelman and Waldfogel 1999).
Our empirical setting, focusing on vertical exchange relationships at a single law firm, effectively
eliminates several other potential sources of unobserved heterogeneity (and possible selection bias) such as
those associated with differences in legal standards, lawyer preferences or reputation. Moreover, the likelihood
of litigation of the disputes in our sample is not significantly related to whether the law firm represented the
plaintiff or the defendant (48 and 54 cases respectively), whether the firm had its own internal legal department
(true in 15 cases), the nationality of the firms involved, or whether the dispute involved disagreement about the
intended scope of the agreement versus alleged failure to meet contract terms such as payment or delivery
schedules (33 and 69 cases respectively).4
The contracts included in our study encompass a variety of vertical exchange relationships and involve
firms in a broad range of industries. The majority of the firms in the sample (53%) are in the manufacturing
sector, with industrial and commercial machinery being the most frequently-represented industry at the 2-digit
4 Adding control variables for these different features produces no material change in our empirical results.
13
SIC level (SIC 35: 14%); 12% are in chemicals and allied products (SIC 28); and 9% in Electronic & Electrical
Equipment & Components (SIC 36); no other single 2-digit manufacturing industry accounts for more than 5% of
the total sample. Retail firms are also well-represented (14%), as are other service firms (33%).
Table 1 provides details of the dispute resolution mode observed in each case in the sample and gives a
breakdown across four types of exchange relationships: distribution agreements, plus three types of supply
agreements: (i) intermediate products and components used in production, (ii) information technology,
software development and implementation; (iii) other services, primarily consulting. Perhaps not surprisingly,
given the involvement of lawyers, many of the disputes do end in litigation. However, close to half of the cases –
41 out of 102 – are settled via private dispute resolution processes: 13 through arbitration, 5 in mediation, and
23 through negotiated settlement. This reinforces the notion that there are strong incentives to settle even
quite serious disputes via private dispute resolution processes.
[Table 1 about here]
Our theoretical arguments regarding the relative attractiveness of private dispute resolution versus
litigation are based in part on the assumption that private dispute resolution procedures are more likely to
uncover zones of hidden agreement between the partners, generate more nuanced settlements, and preserve
the exchange relationship beyond the dispute – in other words that they reflect partner firms’ efforts to “work it
out” when a serious dispute emerges. To probe this assumption directly we examined the terms of the
settlements / judgments recorded in the legal files for each of the disputes, coding them according to the
presence or absence of several features that we believe capture the flexibility and forward-looking aspects (or
lack thereof) of the dispute resolution outcomes. Table 2 displays frequency statistics for the presence of each
of these different terms in cases resolved under each of the different dispute resolution modes in our sample.
[Table 2 about here]
These statistics are quite consistent with our assumption: when compared with each of the private
dispute settlement modes, a judgment resulting from litigation is more likely to include an award of monetary
14
damages to one of the parties but less likely to specify contract modifications to support completion of the
contracted activities, or a commitment to continue performance under the terms of the existing contract. In
addition, in about 50% of the privately-resolved cases the firms indicated their intent to continue with the
exchange into the future, while for those cases that ended in a litigated judgment, only 12% of the parties
indicated a willingness to continue with the relationship.5 Thus private dispute resolution processes indeed
appear to be associated with more nuanced settlements tailored to the changing needs of the contracting
parties, and with greater continuity in exchange relationships.
2.2. Dependent Variable: Dispute Resolution Modes
In keeping with our focus on the choice between litigation and private dispute resolution, in our
empirical estimations we collapse arbitration, mediation, and negotiation into a single category and use a
binomial probit model to estimate the probability of a litigated outcome versus private dispute resolution. Our
dependent variable in the empirical analysis is thus an indicator variable, Litigation which takes the value of 1
where a dispute resulted in litigation and a court-rendered judgment; 0 if the dispute was settled by private
negotiation, mediation or arbitration.6
2.3. Independent Variables
From the legal files and other archival sources, we construct a variety of explanatory and control
variables for our empirical analysis. To test hypothesis H1, we require a measure that captures the technical
complexity of the exchange transaction under dispute. Since we are unable to directly observe the technical
details of the products or services being exchanged in the focal relationship, we rely instead on the extent of the
5 See Malhotra and Lumineau (2011) for further discussion of circumstances affecting the likelihood of continued
collaboration. 6 Multinomial logit regressions, which allow for separate categories for litigation, negotiation/mediation, and arbitration,
(results not shown, available from the authors on request) support the dichotomous treatment and produce estimates
consistent with our reported results. Note also that, consistent with prior research (Siegelman and Waldfogel 1999), the
dispute resolution modes shown in Table 1 are reflective of the final outcome of dispute resolution; in a few cases
settlement occurred after litigation had been initiated. Further analysis of the available data (also not shown) did not reveal
any significant differences related to the timing of settlement.
15
technical specifications attached to the contract governing the exchange.7 Technical Complexity is equal to the
number of pages of technical specifications and we include both the main effect and a square term for Technical
Complexity to capture the hypothesized non-linear relationship between the technical complexity of the
exchange and the likelihood of litigation.
For relationship duration measures related to past and future exchange, we searched the legal files to
establish the following dates: (i) the start date of the first contract linking the exchange partners involved in the
dispute (for first-time transactors this is the current contract); (ii) the date that the dispute arose, as indicated by
the date of the first communication between the firms related to the dispute and recorded in the case file –
usually either an e-mail or a letter from a manager at one of the firms to his/her counterpart at the other firm;
(iii) the projected termination date of the current contract – i.e. the date at which completion of the work was
expected at the time the parties entered the contract, and/or the date at which the agreement was scheduled
to terminate. These dates were then used to construct the following variables:
Pre-Dispute Relationship Duration is equal to the elapsed time (in days) between the beginning of the
exchange relationship and the date at which the dispute arose (as defined above).8 Time Remaining is equal to
the number of days remaining in the existing contract at the time that the dispute arose.9 Since this latter
variable is only defined for contracts with a specified duration, we also include a dummy variable, Open, which is
equal to one for those contracts that are open-ended and have no recorded end date.10
7 This measure resonates with Simon’s (1962, p. 468) definition of complex systems as systems “made up of a large number
of parts that interact in a non-simple way.” The relationship between the extent of technical detail included in the contract and more direct indicators of the underlying technical complexity of the transaction has been demonstrated empirically in
prior work, for example by Argyres et al. (2007) and Vanneste and Puranam (2010). 8 There was one contract in the sample for which we could not find a start date; we set pre-dispute relationship duration to
the sample mean value of 627 days; our results are insensitive to the exclusion of this observation. 9 Time Remaining may understate the parties’ expectations regarding future exchange, in the event that they anticipate
contract renewal and/or the initiation of additional contracts. This represents a limitation of the current study and an
opportunity for future work (see p. 25, below). 10
Note that the construction of these variables means that for a given observation it can never be the case that both Time
Remaining and Open take on a positive value: Open-ended relationships take a value 1 on Open and a value 0 on Time
Remaining; Time-bound contracts take a value of 0 on Open and a value between 0 and 2575 days on Time Remaining,
16
To evaluate the presence of cooperative norms in the exchange relationship prior to the dispute we
applied content analysis and, following prior research (Heide and John 1990; Jap and Ganesan 2000), coded all
descriptions of the exchange relationship in lawyers’ notes from their initial meetings with managers of the
client firm.11
Based on this coding we developed a dichotomous measure: Cooperative Norms = 1 if the file
contains explicit references to flexibility, participation, and/or solidarity in prior interactions between the
partners; 0 otherwise.
We also include a range of control variables that, based on prior research, we anticipate may influence
the likelihood of private dispute resolution. Transaction cost economics for example, suggests that exchange
relationships vary in the extent of specific investments – i.e. investments whose value would be lost or degraded
should the relationship be terminated (Williamson 1991). The build-up of such investments may increase firms’
preference for private dispute resolution procedures in order to preserve the exchange relationship. Although
we are unable to obtain direct measures of specific investments, we again used content analysis of the legal files
and coded for references to relationship-specific assets in the following categories: (a) human assets, i.e.
knowledge specialized to a particular buyer or supplier’s products, or time and effort dedicated to learning
about a buyer’s specific requirements; (b) physical assets, i.e., specialized production equipment and systems
linking buyer and supplier production and scheduling activities; (c) site specificity , i.e. investments in dedicated
facilities specific to this relationship. Given the coarse grain of this categorical data, we follow prior research in
this domain (David and Han 2004, p. 47) and use a simple dichotomous variable, High Asset Specificity = 1 if
there are indications of specific investments in two or more categories; 0 otherwise.12
based on the number of days left in the current contract. This construction facilitates interpretation: a negative coefficient
on Open gives an indication of the reduced likelihood of litigation for open-ended contracts relative to time-bound
contracts with zero time remaining; the negative coefficient on Time Remaining gives an indication of the reduced
likelihood of litigation associated with additional days remaining on time-bound contracts. 11
See Malhotra and Lumineau (2011) for details of the content analysis used in the construction of this variable. 12
The information in the files is not sufficiently nuanced to develop measures of the intensity of investments in each
category; we are more confident that we are able to capture the presence or absence of specific investments of the
17
Given the substantial fixed costs of litigation, prior research suggests that private settlement may also
be more likely for disputes involving contracts of relatively low monetary value (Priest and Klein 1984), where
neither firm has a large stake in the outcome. We thus include a control variable, Contract Value, defined as the
log of the total value of the contract under dispute, in thousands of inflation-adjusted Euros.13
To also account
for the possibility that larger firms are more able to absorb the costs of litigation we include Total Size, the
combined size of the two firms, measured as the sum of the total assets (data obtained from Bureau van Dijk’s
ORBIS database). Similarly, to capture asymmetry in the parties’ stakes in the dispute or the potential impact of
a judgment on their financial health (controlling for contract value), we include an additional variable,
Asymmetry, defined as the log of the absolute value of the difference in revenues of the firms involved.
Finally, a continuous time trend variable, Time Trend, captures possible changes in litigation practices
over the sample period, and dummy variables indicating agreement types (corresponding to those shown in
Table 1) and industry affiliation of the exchange partners are also included. Table 3 shows descriptive statistics
for all of these variables, plus supplementary variables used in robustness tests (described below, following
presentation of the empirical results).
[Table 3 about here]
3. Empirical Results
Our first estimation results, relating to hypothesis H1, are shown in Table 4, Model 1. The dependent
variable in this binomial probit regression is the dichotomous variable, Litigation and a positive coefficient
indicates an increased likelihood of a litigated outcome relative to private dispute resolution.
[Table 4 about here]
different types, but we cannot rule out the possibility that intense investment in some categories dominates the effect of
lower investments in other categories. 13
Ideally we would like to measure each party’s direct stake in the dispute, i.e. the dollar value of that part of the contract
that is under dispute (which may be different than the total contract value). However, because lawsuits sometimes include
non-monetary claims it is not possible to subscribe a direct monetary value to each firm’s stake in the dispute, and contract
value represents the closest available approximation.
18
Looking at the results for Technical Complexity, we see the main effect is positive and the square term is
negative; calculation of the net effect of changes in Technical Complexity and (Technical Complexity)2 on the
probability of litigation (with all other independent variables held at their sample means) reveals an inflexion
point quite low in the observed data range, consistent with the hypothesized inverted-U shape. Figure 1 displays
this relationship graphically. As predicted, we see a relatively low predicted likelihood of litigation for the
technically simplest exchanges (less than 50% for exchanges with zero or very few pages of the contract devoted
to technical specifications), but this rises quite rapidly as complexity increases, to a high in excess of 80%;
beyond the inflexion point the probability of litigation approaches zero for the most technically complex
transactions. This is consistent with the argument that exchanges involving a high degree of technical complexity
predictably tax judges’ ability to reach an efficient resolution of the dispute; recognizing this, partners in such
exchanges eschew the court in favor of private dispute resolution via negotiation or arbitration.14
[Figure 1]
Very few of the control variables in the regression specifications in Table 4 yield significant coefficients.
Among the industry dummies (coefficients suppressed for space considerations), disputes involving firms in
machinery and in “other manufacturing” are significantly less likely to be settled privately relative to the
baseline category, other services; none of the other industry dummies or indicators of agreement type carry
significant coefficients.
The results of regressions examining how dispute resolution mode varies with the elapsed duration of
the exchange relationship are presented in Models 2-4, also in Table 4. In the simplest specification (Model 2)
14 In supplementary analysis (results not shown; available from the authors on request) we relaxed the functional form
restriction imposed by the quadratic formulation, replacing the count variables with a series of piece-wise dummy variables
based on quintiles in the distribution of Technical Complexity, and running a regression equivalent to that in Model 1
(without a square term). The results are quite consistent with those derived from the quadratic specification, yielding
significant negative coefficients on the first and fifth quintiles; coefficients in the three middle quintiles are insignificantly
different from each other. These results thus indicate that litigation is significantly less likely at the two extremes of the
distribution of technical complexity, consistent with H1.
19
the coefficient on Pre-Dispute Relationship Duration is very small in magnitude and statistically insignificant.
Thus we do not find support for the simple effect of elapsed duration on exchange partners’ willingness to work
things out predicted in hypothesis H2. In Model 3 we add the indicator variable Cooperative Norms and see that
disputes arising in exchange relationships characterized by cooperative norms are indeed less likely to be
litigated and tend to be resolved via private dispute resolution procedures, as one would expect. Thus the
development of cooperative norms over the course of exchange appears to significantly tip the balance in favor
of private dispute resolution, allowing firms to avoid costly litigation. Model 4 probes this result further,
interacting Cooperative Norms with Pre-Dispute Relationship Duration. This estimation reveals an interesting
pattern: the main effect of Pre-Dispute Relationship duration is now significant and positive, while Cooperative
Norms interacts negatively with duration, supporting hypothesis H3.15
This result suggests that the impact of
exchange duration on parties’ willingness to work things out when disputes arise is contingent on the
development of norms of cooperation; in the event that such norms do not develop, the probability of a
litigated outcome is actually increased. One possible explanation for this result is that parties foresee little
prospect of successful negotiation or effective compromise within the context of private dispute settlement
procedures when they have failed to develop cooperative norms over the course of an extended exchange
relationship.
[Table 5 about here]
Our next set of results explores the impact of the shadow of the future on firms’ willingness to work
things out. In Table 5, the first regression (Model 5) adds Time Remaining and Open to the basic Model 1
specification. Consistent with hypothesis H4, the coefficients on both Time Remaining and Open are significant
15 We also calculated and graphed the magnitude and significance of the interaction term across all observations using the
“inteff” command in Stata (Norton et al. 2004). These results (available on request) confirmed that the sign of the
interaction effect is indeed negative across the entire range of the data and the effect is significant for observations across
approximately 80% of the data range – as one would expect, the significance of the interaction term is lowest at the
extremes of the data range.
20
and negative indicating that firms are less likely to litigate disputes when the shadow of the future is long.
Models 6 and 7 reintroduce the variables related to pre-dispute duration. The significant effects for the variables
associated with the shadow of the future are robust to the inclusion of these additional variables and again we
see that the impact of prior exchange on the likelihood of litigation is only significant when cooperative norms
have developed over the course of an extended exchange relationship.
4. Robustness Tests
The empirical results reported above suggest that the likelihood that exchange partners end up in court
when they get into a dispute changes significantly depending on their past exchange relationship and
expectations re future exchange as well as on the complexity of the underlying transaction. In this section we
probe the robustness of these results and test some of the underlying assumptions in our analysis.
The primary target of our robustness tests is the implied ceteris paribus condition in each of our
hypotheses. As mentioned earlier, a particularly salient concern in our empirical setting is the inherent selection
bias in our sample. This stems from the fact that we do not observe a disputed exchange until the point at which
lawyers become involved in dispute resolution. This implies that we are only observing a portion of the
distribution of exchanges that are “at risk” of litigation. Since there may be sources of unobserved heterogeneity
in the underlying exchanges that are correlated with our measured attributes as well as with the likelihood of
litigation, the resulting selection bias could interfere with our ability to draw valid empirical inferences from the
analysis (Hamilton and Nickerson 2003). In an ideal world, we would identify a series of instruments that are not
correlated with the likelihood of litigation versus private dispute resolution except indirectly through the
endogenous exchange and relationship constructs of interest (e.g., Technical Complexity, Cooperative Norms).
Unfortunately, as in many other exchange contexts, identifying valid instruments in our setting is infeasible,
given that we observe no relevant exogenous shocks during the sample period.
Even in the absence of valid instruments, we are nonetheless able to address some of the most likely
effects of problematic bias in our sample, at least indirectly. Take, for example, the impact of Pre-Dispute
21
Relationship Duration on the likelihood of litigation and, in particular, the observed positive association between
the likelihood of litigation and the length of prior exchange when cooperative norms have not developed (i.e.
when Cooperative Norms = 0). An alternative explanation for this observation is that lawyers may not become
involved as early in the process in conflicts between repeated exchange partners and, since we only observe
disputes once lawyers do get involved, the disputes between partners with a longer history of exchange in our
sample may be on average more “serious” (in a way that is unobservable to us) than those involving first-time or
recent exchange partners. To assess this possibility we coded an additional variable, Time to Lawyer, equal to
the number of days separating the first communication between the firms themselves regarding the dispute,
and the first meeting between the lawyer and the client firm about the dispute. Table 6 shows relevant
summary statistics for disputes between firms with different relational histories.
[Table 6 about here]
These statistics are quite illuminating: for the sample as a whole, there is little difference in the mean
value of Time to Lawyer between exchange relationships where Cooperative Norms have developed and those
where they have not (161 versus 145 days; not a significant difference, given the sample variance). However, for
firms that have been in an exchange relationship for a long time (>800 days, the 75th
percentile in our sample)
the difference in Time to Lawyer for exchange relationships with cooperative norms versus those without is
dramatic: 236 days versus 36 days. This strongly reinforces our inference that partners with a long history of
cooperation in an exchange relationship have a higher willingness to “work it out” to avoid litigation than do
either new exchange partners or those with a troubled exchange history: Partners with a long history of
cooperative exchange try to work things out between themselves for a much longer period before involving
lawyers; moreover, once lawyers do get involved, these long-time exchange partners are still more likely to
eschew litigation and resolve the dispute privately. For partners with an extended negative experience of
exchange (i.e., where cooperative norms have failed to develop) this process is completely reversed: lawyers
22
become involved very early in the process – much earlier even than is the case for new exchange partners – and
litigation is the more likely outcome in this case.
Another possible source of unobserved heterogeneity (and problematic endogeneity bias) in our sample
is differential contracting capability, wherein some exchange partners are better-able to accurately foresee
potential sources of conflict, and address these by adding appropriate additional clauses to the contract
governing the exchange. This possibility is particularly problematic to the extent that there is “learning by
contracting,” as suggested in some prior research (e.g., Mayer and Argyres 2004); in this case one might observe
a spurious negative correlation between Pre-Dispute Relationship Duration and the likelihood of litigation. To
probe for this possibility we examined the contracts used to govern the exchanges in our sample and created a
count variable based on the presence or absence of nine particular clauses in the contract, similar to those used
in prior research (e.g., Parkhe 1993, Lumineau and Malhotra 2011).16
The index variable, Contractual
Sophistication is defined as: ∑Ci; where Ci=1 if provision i exists; Ci=0 otherwise. This is an integer variable
ranging from zero to nine.
Some contracts in our sample also include an explicit dispute resolution provision. Inclusion of such a
provision may itself reflect an increased commitment to ‘working it out’ in the event of a dispute, although as
legal scholars emphasize, it by no means rules out the possibility of litigation (or any other dispute resolution
mode): “Even if parties have contractually agreed to use one *dispute resolution+ method, they may switch to
another if they feel that the latter is more appropriate for a given dispute” Mattli (2001, p. 920). We should
nonetheless ensure that our results regarding final dispute resolution outcomes are robust to the inclusion of
16 The contract clauses included in the measure are: 1. Right to audit/inspection; 2. Safeguard system; 3. Control by a third
party; 4. Penalty clause; 5. Resolution clause; 6. Assignment of roles and responsibilities; 7. Indication of duration and
conditions of renewal; 8. Organizational coordination (ability to reassign tasks among participants without altering the goal
of the contractual arrangement); and 9. Strategic coordination (process set up to redefine the objective of the relationship).
Initial coding of the contracts was undertaken with input from a law professor and three practicing lawyers specializing in
contract law (with no connection to the disputes or to the law firm providing the data); measures were further validated by
six other contract law specialists.
23
this clause, and thus introduce a dummy variable Dispute Resolution Clause indicating the presence of such a
clause in the contract. Regression results incorporating these contract-based measures are shown in Table 7. As
these results indicate, both the inclusion of a dispute resolution clause and increases in contractual
sophistication are associated with decreased likelihood of litigation (in most specifications), but inclusion of
these terms does not materially change the main results noted earlier.
[Table 7 about here]
5. Discussion and Conclusion
The empirical analysis presented in this paper brings together previously disparate literatures in law and
economics, organizational economics and relational exchange to provide a first empirical examination of dispute
resolution under the imminent threat of litigation in vertical exchange relationships. Our unique data allow us to
observe the results of firms’ attempts to “work it out” when serious contract disputes arise and to explore the
choice of dispute resolution procedures under different exchange conditions. The picture that emerges from our
analysis is consistent with prior literature on the governance of inter-firm exchange; our study also extends this
prior literature and contributes to broader debates on firm boundaries and inter-organizational exchange.
First, our study serves to counter the received wisdom that the involvement of lawyers necessarily
signals the bitter end of an exchange relationship. We see that even in the event that quite serious disputes
emerge and lawyers get involved, firms manage to avoid litigation and resolve their disputes privately in many
cases. This is particularly true for technically very simple exchanges, where private settlement is relatively
straightforward and avoids the fixed costs and delays associated with litigation; it is also true for the most
complex of exchanges where firms appear to eschew the courts altogether because they realize that court
proceedings are ill-equipped to handle the nuances involved with technically-complex disputes, and firms
instead choose a private forum for dispute resolution.
Second, our study contributes to ongoing debates about the impact of relational history on exchange
partners’ behavior. Here we see that a longer history of exchange does not automatically lead to an increased
24
willingness to work things out when serious disputes arise (even allowing for the possibility that fewer such
disputes reach the stage where lawyers get involved). Longer pre-dispute duration has little impact on the
likelihood of litigation in itself; only when cooperative norms develop is litigation less likely, and this effect is
strongest when such norms build up over the course of extended exchange. In addition to this link between
dispute resolution and the “shadow of the past,” we also observe that the “shadow of the future” has a
significant effect on the likelihood of litigation, suggesting that firms are indeed forward-looking in their
assessments of the costs and consequences of different dispute resolution procedures. As this last point
indicates, the results of our study are suggestive of a “discriminating alignment” between exchange
characteristics and the choice of dispute resolution procedure, in keeping with the spirit of Williamson’s (1991)
arguments about the legal underpinnings of different forms of governance.
Our study of course has limitations, the most important being the selection bias inherent in our sample
since we do not observe conflicts between exchange partners unless they escalate to the point where lawyers
become involved. However, as our supplementary analysis of the timing of lawyer involvement suggests, the
most likely bias only strengthens our inference that partners in long-standing cooperative exchange
relationships are indeed more able to work things out in private dispute resolution procedures. We also provide
evidence that, while the contracts governing the exchange relationships found in our sample are consistent with
efforts to reduce the likelihood of litigation through the inclusion of additional contract terms, this does not
substitute for the effect of past or anticipated future exchange.
Another limitation of our study relates to its setting in the French civil law system: although a significant
fraction of the firms involved in the contracts in our sample are from outside France all of the contracts are
subject to French contract law, and litigation, when it occurs, takes place in French courts. Discussions with legal
experts lead us to believe that, for the purposes of our research, there are no material differences in the legal
rules governing contracts or dispute resolution in the French civil law system from those in the US common law
system (Deffains and Kirat 2001). However, accounts of some differences in the propensity to enter mediation
25
and in the costs and delays associated with litigation suggest that the incentives to settle disputes privately may
be even higher in the US system than is the case in our setting (Yelpaala et al. 1986). Thus, although we believe
that our results are not dependent on idiosyncratic features of the French legal system, it would be very
interesting to replicate the study in a different legal context. Indeed, comparing dispute resolution processes
across institutional contexts represents an intriguing avenue for future research; a multidisciplinary approach to
this line of research could be particularly worthwhile, as the most insightful results are likely to emerge from
studies that engage both the legal and institutional rules that guide dispute resolution, and the organizational
and cultural contexts in which the disputed exchange relationships take place.
There are also several additional data items that could facilitate refinement and further extension of our
study. With more direct measures of the scope and complexity of the actual products and services exchanged,
one could avoid a significant simplification in the current analysis, which equates technically-complex contracts
with complexity in the products and services being exchanged. Similarly, our measure of asset specificity is quite
crude and imprecise – a fact which may account for the general lack of significance of this variable in our
empirical analysis. Our view of the ‘shadow of the future’ is also limited to the time horizon associated with the
current contract; we cannot observe the extent to which firms foresee additional exchange opportunities in the
future. Better measures would likely strengthen our observed association between the length of the shadow of
the future and the likelihood of private dispute resolution.
In sum, while we do not believe that data limitations significantly undermine the contributions of our
study, we are conservative in the empirical inferences that we draw and, in particular, stop short of claiming any
causal link between exchange features and dispute resolution outcomes. Further disentangling of the
relationships among exchange content, relational context, contractual governance and dispute resolution mode
represents an important avenue for continuing research, albeit one that poses significant data challenges. A
research program along these lines holds the potential to provide answers to questions that continue to
generate interest and disagreement among organizational and legal scholars, such as: what are the critical
26
organizational and institutional supports for relational governance? To what extent does relational governance
operate “in the shadow of the law” in inter-firm exchange relationships? What does it really mean for exchange
partners to adopt the concept of “contract as framework” (Llewellyn 1931)? How do competitive dynamics
within an industry impact firms’ legal strategies (including contract dispute resolution), and how do these legal
strategies, in turn, shape other organizational choices? While we are a long way from establishing definitive
answers to these questions, we believe that our study injects important new evidence into the debates.
In reaching outside of the usual domain of organizational and management research to leverage findings
from law and economics we believe that, not only have we introduced a new perspective on dispute resolution
in exchange relationships, but also highlighted the potential for further collaboration between these disciplines.
We invite others to explore opportunities in multidisciplinary research to illuminate different aspects of
organization and exchange.
27
References
Anderson, E., S. Jap. 2005. The dark side of close relationships. MIT Sloan Management Rev. 46(3) 75–82.
Argyres, N. S., J. Bercovitz, K. J. Mayer. 2007. Complementarity and evolution of contractual provisions: An
empirical study of IT service contracts. Organ. Sci. 18(1) 3–19.
Ariño, A., J. de la Torre. 1998. Learning from failure: Towards an evolutionary model of collaborative ventures.
Organ. Sci. 9(3) 306–325.
Axelrod, R. 1984. The Evolution of Cooperation. Basic Books, New York.
Bebchuk, L. A. 1984. Litigation and settlement under imperfect information. RAND J. Econom. 15(3) 404–415.
Bouwens, J., L. van Lent. 2006. Performance measure properties and the effect of incentive contracts. J.
Management Accounting Res. 18 55–75.
Carson, S. J., A. Madhok, T. Wu. 2006. Uncertainty, opportunism, and governance: The effects of volatility and
ambiguity on formal and relational contracting. Acad. Management J. 49(5): 1058–1077.
Daft, R. L., N. B. Macintosh. 1981. A tentative exploration into the amount and equivocality of information
processing in organizational work units. Admin. Sci. Quart. 26 207–225.
David, R. J., S-K Han. 2004. A systematic assessment of the empirical support for transaction cost economics.
Strategic Management J. 25 39–58.
Deffains, B., T. Kirat. 2001. Law and Economics in Civil Law Countries. Elsevier JAI Press, Amsterdam.
Dyer, J., H. Singh. 1998. The relational view: Cooperative strategy and sources of interorganizational competitive
advantage. Acad. Management Rev. 23 660–679.
Farrell, J., E. Maskin. 1989. Renegotiation in repeated games. Games and Economic Behavior 1 327–360.
Golden, B. R. 1992. The past is the past – Or is it? The use of retrospective accounts as indicators of past
strategy. Acad. Management J. 35(4) 848–860.
Gould, J. P. 1973. The economics of legal conflicts. J. Legal Stud. 2 279–300.
Grossman, S., O. Hart. 1986. The costs and benefits of ownership: A theory of vertical and lateral integration. J.
Political Econom. 94 691–719.
Gulati, R. 1995. Does familiarity breed trust? The implications of repeated ties for contractual choice in alliances.
Acad. Management Rev. 38 85–112.
Gulati, R., M. Sytch. 2008. Does familiarity breed trust? Revisiting the antecedents of trust. Manage. Decis. Econ.
29 165–190.
Hamilton, B., J. A. Nickerson. 2003. Correcting for endogeneity in strategic management research. Strategic
Organ. 1 53–80.
Heide, J. B., G. John. 1990. Alliances in industrial purchasing: The determinants of joint action in buyer supplier
relationships. J. Marketing Res. 27(1) 24–36.
Heide, J. B., A. Miner. 1992. The shadow of the future: Effects of anticipated interaction and frequency of
contact on buyer-supplier cooperation. Acad. Management J. 35(2) 265–291.
Hoopes, D., S. Postrel. 1999. Shared knowledge, “glitches” and product development performance. Strategic
Management J. 20 837–852.
28
Jap, S. D., S. Ganesan. 2000. Control mechanisms and the relationship life cycle: Implications for safeguarding
specific investments and developing commitment. J. Marketing Res. 37(May) 227–245.
Johnson, S., J. McMillan, C. Woodruff. 2002. Courts and relational contracts. J. Law Econom. Organ. 18 221–277.
Kanter, R. M. 1994. Collaborative advantage: The art of alliances. Harvard Business Rev. 72(4) 96–109.
Klein, B. 1996. Why hold-ups occur: The self-enforcing range of contractual relationships. Econom. Inquiry 34
444–463.
Klein Woolthuis, R. K., B. Hillebrand, B. Nooteboom. 2005. Trust, contract and relationship development. Organ.
Stud. 26(6) 813–840.
Knight, F. 1921. Risk, Uncertainty and Profit. Houghton Mifflin, New York.
Lewis, D., A. Weigert. 1985. Trust as a social reality. Soc. Forces 63(4) 967–985.
Liker, J. K., T. Y.Choi. 2004. Building deep supplier relationships. Harvard Bus. Rev. 82(12) 104–113.
Llewellyn, L. 1931. What price contract? An essay in perspective. Yale Law J. 40 704–751.
Lumineau, F., D. Malhotra. 2011. Shadow of the contract: How contract structure shapes interfirm dispute
resolution. Strategic Management J. Forthcoming.
Macaulay, S. 1963. Non-contractual relationships in business: A preliminary study. Amer. Soc. Rev. 28 55–70.
Malhotra, D., F. Lumineau. 2011. Trust and collaboration in the aftermath of conflict: The effects of contract
structure. Acad. Management J. Forthcoming.
Mattli, W. 2001. Private justice in a global economy: From litigation to arbitration. Internat. Organ. 55(4) 919–947.
Mayer, K. J., N. Argyres. 2004. Learning to contract: Evidence from the personal computer industry. Organ. Sci.
15 394–410.
Mohr, J., R. Spekman. 1994. Characteristics of partnership success: Partnership attributes, communication
behavior, and conflict resolution techniques. Strategic Management J. 15(2) 135–142.
Monteverde, K., D. J. Teece. 1982. Appropriable rents and quasi-vertical integration. J. Law Econom. 25(2) 321–328.
Norton, E. C., H. Wang, C. Ai. 2004. Computing interaction effects in logit and probit models. Stata J. 4(2) 103–116.
Park, S. H., G. R. Ungson. 2001. Interfirm rivalry and managerial complexity: A conceptual framework of alliance
failure. Organ. Sci. 12(1) 37–53.
Parkhe, A. 1993. Strategic alliance structuring: A game theoretic and transaction cost examination of interfirm
cooperation. Acad. Management J. 36 794–829.
Poppo L., T. R. Zenger. 2002. Do formal contracts and relational governance function as substitutes or
complements? Strategic Management J. 23(8) 707–726.
Poppo, L., K. Z. Zhou, T. R. Zenger. 2008. Examining the conditional limits of relational governance: Specialized
assets, performance ambiguity and long-standing ties. J. Management Stud. 45 1195–1216.
Priest, G. L., B. Klein. 1984. The selection of disputes for litigation. J. Legal Stud. 13 1–55.
Richman, B. D. 2004. Firms, courts, and reputation mechanisms: Toward a positive theory of private ordering.
Columbia Law Rev. 104 2328–2367.
29
Riskin, L. L., J. E. Westbrook, C. Guthrie, R. Reuben, J. Robbennolt, N. A. Welsh. 2009. Dispute Resolution &
Lawyers, 4th
Edition. St Paul, MN: West Group.
Siegelman, P., J. Waldfogel. 1999. Towards a taxonomy of disputes: New evidence through the prism of the
Priest/Klein model. J. Legal Stud. 28/1 101–130.
Simon, H. A. 1962. The architecture of complexity. Proc. Amer. Phil. Soc. 106(6) 467–482.
Simonin, B. 2007. Navistar cuts off engines to Ford. Financial Times. (February 26).
Slater, G., D. Spencer. 2000. The uncertain foundations of transaction cost economics. J. Econ. Issues 31 61–87.
Tractenberg, P. L. 1984. Training lawyers to be more effective dispute preventers and dispute settlers:
Advocating for non-adversarial skills. J. Dispute Resolution 87–118.
Uzzi, B. 1996. The sources and consequences of embeddedness for the economic performance of organizations:
The network effect. Amer. Sociological Rev. 61 674–698.
Vanneste, B. S., P. Puranam. 2010. Repeated interactions and contractual detail: Identifying the learning effect.
Organ. Sci. 21 186–201.
Williamson, O. E. 1985. The Economic Institutions of Capitalism. Free Press, New York.
Williamson, O. E. 1991. Comparative economic organization: The analysis of discrete structural alternatives.
Admin. Sci. Quart. 36 269–296.
Yelpaala, K., M. Rubino-Sammartano, D. Campbell. 1986. Drafting and Enforcing Contracts in Civil and Common
Law Jurisdictions. Kluwer Law and Taxation, Deventer.
Zollo, M., J. J. Reuer, H. Singh. 2002. Interorganizational routines and performance in strategic alliances. Organ.
Sci. 13(6) 701–713.
________________________________________
Acknowledgments:
The authors are grateful to the members of the law firm for their help and generosity in providing data access,
and to the law professors who provided expert advice and assistance. In addition, we wish to thank senior editor
Nick Argyres and the three Organization Science referees for their constructive feedback; conversations with Eric
Brousseau, Bertrand Quélin and Stéphane Saussier, and comments from seminar and conference participants
too numerous to mention were also enormously helpful; errors remain our responsibility
30
Figure 1 – Change in Probability of Litigation with Increasing Technical Complexity
0
.2.4
.6.8
0 10 20 30Pages of Technical Detail
31
Table 1 – Agreement Types and Dispute Resolution Modes
Number of Cases Ending in…
Litigated
outcome
Arbitration Mediation Negotiated
settlement
Total
Distribution agreement 22 3 2 9 36
Supply agreement –
intermediate products 13 8 1 8 30
Supply agreement – IT 19 1 2 5 27
Supply agreement –
other services 7 1 0 1 9
Total 61 13 5 23 102
Table 2 – Settlement/Judgment Terms in Different Dispute Resolution Modes
Litigated
Outcome Arbitration Mediation
Negotiated
settlement Total
Monetary damages 52
(85.2%)
9
(69.2%)
2
(40%)
15
(65.2%)
78
(76.5%)
Specific performance
based on terms of
existing contract
12
(19.6%)
8
(61.5%)
4
(80%)
17
(73.9%)
41
(40.2%)
Adjustment of
contract terms
4
(6.5%)
4
(30.7%)
2
(40%)
8
(34.7%)
18
(17.6%)
Stated intention to
continue exchange
7
(11.4%)
6
(46.1%)
3
(60%)
13
(56.5%)
29
(28.4%)
Total # of disputes in
sample (n) 61 13 5 23 102
Notes:
1. Numbers in columns do not add up to 100%: a settlement or judgment typically includes multiple terms,
and each cell indicates the frequency of observation of each term in settlements or judgments
associated with each dispute resolution mode.
2. Differences of means tests indicate significant differences between litigated judgments and each of the
three private dispute resolution modes on all dimensions, and no significant differences among
arbitration, mediation and negotiation except for monetary damages, which are more likely with
arbitration.
32
Table 3 – Descriptive Statistics
Variables Mean S.D Range 1 2 3 4 5 6 7 8 9 10 11 12 13
1 Litigation 0.59 0.49 0-1
2 Technical
complexity
9.20 18.63 1-122 -0.25*
3 Pre-dispute
relationship
duration
627.4 946.2 0-5771 0.03 -0.03
4 Time remaining
245.0 551.5 0-2575 -0.52* 0.24* -0.09
5 Open 0.34 0.47 0-1 0.00 0.06 0.18 -0.32*
6 Cooperative
norms
0.54 0.50 0-1 -0.38* 0.15 0.06 0.38* 0.11
7 High asset
specificity
0.35 0.48 0-1 0.01 0.33* -0.07 0.00 -0.18 -0.07
8 Contract value 2.42 0.66 0.70-
4.33
0.09 -0.15 -0.22* 0.06 -0.10 -0.09 0.23*
9 Total size 7.90 0.83 5.85-
10.31
-0.11 0.12 -0.06 -0.02 0.05 -0.02 -0.03 -0.07
10 Asymmetry 7.66 0.96 5.16-
10.28
-0.14 0.14 -0.11 0.01 0.02 -0.01 -0.00 0.00 0.95*
11 Time trend 8.79 3.42 0-14 -0.20* 0.07 -0.26* 0.18 -0.08 0.21* 0.12 0.13 -0.01 0.03
12 Time to lawyer 154.0 158.4 0-854 -0.13 -0.00 0.18 -0.09 0.19* 0.04 -0.16 -0.27* 0.06 -0.03 -0.21*
13 Dispute
resolution clause
0.48 0.50 0-1 -0.25* 0.09 -0.16 -0.04 0.29* 0.04 0.35* 0.15 0.02 0.01 0.19 -0.06
14 Contractual
sophistication
4.87 1.75 0-9 -0.12 0.42* -0.23* 0.26* -0.35* 0.08 0.64* 0.28* 0.00 0.06 0.18 -0.16 0.20*
* p < 0.05
33
Table 4 – Litigation and the Shadow of the Past
Model 1 Model 2 Model 3 Model 4
Technical complexity 0.274 ** 0.256 ** 0.259 ** 0.389 ***
(0.108) (0.103) (0.110) (0.134)
(Technical complexity)2 -0.016 *** -0.015 *** -0.016 *** -0.023 ***
(0.006) (0.005) (0.006) (0.007)
Pre-dispute relationship duration -0.000 -0.000 0.004 ***
(0.000) (0.000) (0.001)
Cooperative norms -1.363 *** -0.166
(0.367) (0.468)
Pre-dispute relationship duration *
Cooperative norms -0.005 ***
(0.001)
High asset specificity 0.327 0.352 0.216 0.343
(0.322) (0.322) (0.340) (0.362)
Contract value 0.227 0.179 0.152 0.210
(0.256) (0.256) (0.314) (0.337)
Total size 0.522 0.634 0.957 0.850
(0.628) (0.661) (0.660) (0.740)
Asymmetry -0.562 -0.698 -1.050 -1.185
(0.559) (0.590) (0.623) (0.732)
Time trend -0.090 * -0.101 * -0.078 -0.093
(0.048) (0.051) (0.052) (0.058)
Constant 1.149 1.674 2.174 3.261
(1.841) (1.792) (1.785) (2.144)
Agreement Types Yes Yes Yes Yes
Industry Affiliations Yes Yes Yes Yes
Wald Chi2 37.06 *** 40.46 *** 44.16 *** 43.21 ***
Pseudo R2 0.308 0.313 0.418 0.509
* p < 0.10; ** p < 0.05; *** p < 0.01; N = 102
34
Table 5 – Litigation and the Shadow of the Future
Model 5 Model 6 Model 7
Technical complexity 0.262 ** 0.256 ** 0.362 **
(0.116) (0.110) (0.121)
(Technical complexity)2 -0.014 *** -0.014 *** -0.020 ***
(0.005) (0.005) (0.006)
Time remaining -0.035 *** -0.036 *** -0.038 ***
(0.009) (0.010) (0.012)
Open -1.558 *** -1.553 *** -1.588 ***
(0.399) (0.404) (0.431)
Pre-dispute relationship duration -0.000 0.004 **
(0.000) (0.001)
Cooperative norms 0.824
(0.672)
Pre-dispute relationship duration * Cooperative
norms
-0.004 **
(0.002)
High asset specificity 0.247 0.303 0.605
(0.511) (0.487) (0.651)
Contract value 0.307 0.241 0.214
(0.383) (0.411) (0.454)
Total size -0.055 -0.001 0.007
(0.595) (0.590) (0.665)
Asymmetry 0.078 0.027 0.061
(0.531) (0.521) (0.612)
Time trend -0.022 -0.028 -0.033
(0.056) (0.057) (0.066)
Constant 0.648 0.914 -0.115
(1.783) (1.812) (2.115)
Agreement Types Yes Yes Yes
Industry Affiliations Yes Yes Yes
Wald Chi2 48.82 *** 47.32 *** 39.88 ***
Pseudo R2 0.600 0.603 0.646
* p < 0.10; ** p < 0.05; *** p < 0.01; N = 102
35
Table 6 – “Time to Lawyer”
Number of days from first communication between the firms themselves regarding the dispute, and the
first meeting between the lawyer and the client firm:
Mean Std Dev Range
Full Sample 160 158 2-854
Cooperative norms=1 161 151 0-854
Cooperative norms=0 145 167 0-786
“Positive” pre-dispute relationship duration > 800
days
236 203 8-854
“Negative” pre-dispute relationship duration > 800
days
36 30 0-98
36
Table 7 – Robustness Tests: Controlling for Contractual Sophistication
Model 8
Model 9
Model 10
Technical complexity 0.391 *** 0.576 *** 4.540 ***
(0.116) (0.166) (1.389)
(Technical complexity)2 -0.021 *** -0.031 *** -0.224 ***
(0.006) (0.008) (0.066)
Contractual sophistication -0.279 ** -0.415 ** -0.589
(0.127) (0.185) (0.561)
Dispute resolution clause -1.230 *** -1.680 *** -18.80 ***
(0.408) (0.549) (5.470)
Time remaining -0.199 ***
(0.054)
Open -6.025 ***
(2.034)
Pre-dispute relationship duration 0.005 ***
(0.001)
Cooperative norms -0.034
(0.439)
Pre-dispute relationship duration *
Cooperative norms
-0.006 ***
(0.001)
High asset specificity 1.152 ** 1.362 ** 6.327 ***
(0.445) (0.549) (1.721)
Contract value 0.600 ** 0.770 3.545 *
(0.283) (0.356) (1.907)
Total size 0.288 0.761 0.710
(0.609) (0.805) (1.601)
Asymmetry -0.294 -1.386 * 3.150 *
(0.534) (0.733) (1.540)
Time trend -0.075 * -0.083 -0.119
(0.047) (0.058) (0.148)
Constant 1.431 6.781 ** -21.35 **
(2.086) (2.425) (6.293)
Agreement Types Yes Yes Yes
Industry Affiliations Yes Yes Yes
Wald Chi2 53.29 *** 48.67 ** 32.12 ***
Pseudo R2 0.402 0.619 0.836
* p < 0.10; ** p < 0.05; *** p < 0.01; N = 102